Arabidopsis CYP735A1 and CYP735A2 encode cytokinin hydroxylases that catalyze the biosynthesis of trans-Zeatin.

Cytokinins (CKs), a group of phytohormones, are adenine derivatives that carry either an isoprene-derived or an aromatic side chain at the N(6) terminus. trans-Zeatin (tZ), an isoprenoid CK, is assumed to play a central physiological role because of its general occurrence and high activity in bioassays. Although hydroxylation of isopentenyladenine-type CKs is a key step of tZ biosynthesis, the catalyzing enzyme has not been characterized yet. Here we demonstrate that CYP735A1 and CYP735A2 are cytochrome P450 monooxygenases (P450s) that catalyze the biosynthesis of tZ. We identified the genes from Arabidopsis using an adenosine phosphate-isopentenyltransferase (AtIPT4)/P450 co-expression system in yeast. Co-expression of AtIPT4 and CYP735A enabled yeast to excrete tZ and the nucleosides to the culture medium. In vitro, both CYP735As preferentially utilized isopentenyladenine nucleotides rather than the nucleoside and free base forms and produced tZ nucleotides but not the cis-isomer. The expression of CYP735A1 and CYP735A2 was differentially regulated in terms of organ specificity and response to CK. Root-specific induction of CYP735A2 expression by CK suggests that the trans-hydroxylation is involved in the regulation of CK metabolism and signaling in roots.

In higher plants, the first step of CK biosynthesis is catalyzed by adenosine phosphate-isopentenyltransferase (IPT), which uses dimethylallyl diphosphate as an isoprene donor to prenylate adenosine phosphate. In Arabidopsis thaliana, seven IPT genes (AtIPT1 and AtIPT3-AtIPT8) have been identified (6,7). Biochemical studies suggest that the AtIPTs preferentially utilize ADP or ATP rather than AMP, resulting in the production of iP riboside 5Ј-diphosphate (iPRDP) or iP riboside 5Ј-triphosphate (iPRTP), respectively (7,8) (see Fig. 1). The low affinity of the AtIPTs for AMP implies that most of the natural iP riboside 5Ј-monophosphate (iPRMP) is formed by dephosphorylation of iPRDP and iPRTP, phosphorylation of iP riboside (iPR) by adenosine kinase, and conjugation of a phosphoribosyl moiety to iP by adenine phosphoribosyltransferase (3) (see Fig. 1). The nucleotides are further dephosphorylated and deribosylated resulting in iPR and iP, respectively. A series of modifications of the adenine moiety of CKs occurs by reactions that are part of a purine metabolic pathway (3).
Classical bioassays (4,5) and recent analyses of CK receptors in Arabidopsis (9,10) and maize (11) demonstrated that free base CKs, such as tZ and iP, are physiologically active forms. Moreover, differential ligand preferences of the receptors (11) imply that side chain variations in isoprenoid CKs are highly significant for the diversification of the hormones' biological function.
Biochemical and genetic mechanisms that control side chain variation have not been characterized well yet. At present, it is assumed that tZ could be synthesized via the iPRMP-dependent and the iPRMP-independent pathways (12) (see Fig. 1). In the iPRMP-dependent pathway, tZ is formed by hydroxylation of iP-type CKs. The enzymatic activity hydroxylating the prenyl side chain was detected in a microsomal fraction of cauliflower (13). It catalyzed the conversion of iP and iPR to tZ and tZ riboside (tZR), respectively; reactivities acting on the nucleotides were not examined (13). Sensitivity to metyrapone, an inhibitor of cytochrome P450 monooxygenase (P450), and dependence of the activity on NADPH suggested that the reaction was catalyzed by a P450 (13). However, this enzyme has not been further characterized.
In the iPRMP-independent pathway, tZR 5Ј-phosphates are proposed to be produced directly by IPT using an unknown hydroxylated side chain precursor that may be derived from the cytosolic mevalonate (MVA) pathway (12) (Fig. 1). The synthesis of tZR 5Ј-monophosphate (tZRMP) from AMP and 4-hydroxy-3-methyl-2-(E)-butenyl diphosphate (HMBDP), a likely precursor, has been demonstrated to be catalyzed by TZS, an agrobacterial IPT (14). However, HMBDP was identified as a metabolic intermediate of the methylerythritol phosphate (MEP) pathway (15), which occurs in bacteria and plastids (16). Selective labeling experiments using 13 Clabeled precursors specific for either the MEP or MVA pathway demonstrated that the isoprenoid side chain of iP and tZ predominantly originates from the MEP pathway, whereas a large fraction of the cZ side chain is derived from the MVA pathway (17). However, the utilizing efficiency of Arabidopsis IPT was much lower for HMBDP than for dimethylallyl diphosphate in vitro (18). Thus, the physiological significance of the iPRMP-independent pathway remains obscure to date.
Although several routes might contribute to tZ production in higher plants, the physiological importance and functional differentiation of the possible pathways is unclear. As a first step toward an understanding of the trans-hydroxylation step of iP-type CKs in the iPRMP-dependent pathway, we attempted to identify CK hydroxylase genes in Arabidopsis. Here we demonstrate that CYP735A encode CK trans-hydroxylases that catalyze the biosynthesis of tZ and discuss the physiological role of the trans-hydroxylation in the regulation of CK metabolism and signaling.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-cDNA for an Arabidopsis NADPH-P450 reductase (ATR1) (19) was synthesized by reverse transcriptase-PCR and cloned into the NotI/SpeI site of the pESC-LEU vector (Stratagene) to yield a plasmid designated pESC-ATR1. The coding region of AtIPT4 was ligated into the ApaI/SalI site of pESC-ATR1 to yield a plasmid designated pESC-ATR1-IPT4. For preparation of the coding region of various P450s, we designed PCR primers based on predicted sequences from the Arabidopsis P450 data base (www.biobase.dk/P450/index. shtml). Each of the reverse transcriptase-PCR products was ligated into the pYES2.1/V5-His-TOPO vector (Invitrogen) to yield a series of plasmids designated pYES-P450. pYES2.1/V5-His/lacZ (Invitrogen) was used as a control plasmid (pYES-lacZ).
Screening of CK Hydroxylase Genes-pESC-ATR1-IPT4 and each member of the pYES-P450 series were co-transformed into Saccharomyces cerevisiae strain YPH499 (Stratagene). The yeast cells were grown in complete minimal drop-out medium (20) without leucine and uracil (CM-Leu-Ura) and with 1% raffinose at 30°C for 24 h. The liquid medium was changed to CM-Leu-Ura with 2% galactose, 1% raffinose, and 80 g/ml 5-aminolevulinic acid, and the A 600 was adjusted to 0.4 and further cultured overnight at 20°C. After removal of cells by centrifugation, the supernatant was mixed with an equal volume of 2% acetic acid. The CK content was analyzed by liquid chromatographymass spectrometry (LC-MS).
CK Analysis-CKs were analyzed with a LC-MS system (model 2695/

FIG. 1. Current model of tZ-type CK biosynthesis in
Arabidopsis via the iPRMP-dependent and the iPRMP-independent pathways. Isoprenoid side chain of iP and tZ predominantly originates from the MEP pathway, whereas a large fraction of the cZ side chain is derived from the MVA pathway (green arrows). Plant IPTs preferably utilize ATP or ADP as isoprenoid acceptors to form iPRTP and iPRDP, respectively (blue arrows). Dephosphorylation of iPRTP, phosphorylation of iPR by adenosine kinase (AK) or conjugation of phosphoribosyl moieties to iP by adenine phosphoribosyltransferase (APRT) create the metabolic pool of iPRMP and iPRDP. This study reveals that the CK nucleotides are converted to the corresponding tZ nucleotides by CYP735As (red arrows). iP, tZ, and the nucleosides could be catabolized by CKX to adenine (Ade) or adenosine (Ado). cZ and tZ can be enzymatically interconverted. tZ could be reversibly converted to the O-glucoside by zeatin O-glucosyltransferase (ZOG) and ␤-glucosidase (␤Glc). The width of the arrowheads and lines in the green, blue, and red arrows indicates the strength of metabolic flow. Flows indicated by black arrows are not well characterized to date. tZRDP, tZR 5Ј-diphosphate; tZRTP, tZR 5Ј-triphosphate. ZQ2000MS; Waters) on a reverse-phase column (Symmetry C 18 , 5 m, 2.1 mm ϫ 150 mm; Waters) at a flow rate of 0.25 ml/min in 0.1% acetic acid with a linear gradient of methanol (0% for 1 min, 0 -50% for 15 min, and 50 -70% for 6 min). CKs were quantified in the selected ion recording mode. Monitored ions for each CK species were as follows: m/z 220 and 352 for tZR and cZ riboside, m/z 136 and 220 for tZ and cZ, m/z 222 and 354 for DZ riboside (DZR), m/z 136 and 222 for DZ, m/z 204 and 336 for iPR, m/z 136 and 204 for iP, and m/z 357 for [ 2 H 5 ]tZR. Other conditions for MS analysis were as described previously (8).
Preparation of Microsomal Fraction-Yeast cells harboring pESC-ATR1 and pYES-CYP735As were cultured in 500 ml of medium under inductive conditions as described above. The microsomal fraction was prepared by the method of Venkateswarlu et al. (21), except that the buffer A did not contain reduced glutathione. After ultracentrifugation, the pellet was further washed once with the buffer A. The final preparation of microsomal pellet was resuspended in the buffer A. Protein concentration was determined with protein assay (Bio-Rad) using bovine serum albumin as the standard. The P450 content was estimated by the method of Omura and Sato (22).
Synthesis of iP Nucleotides-iPRMP, iPRDP, and iPRTP were synthesized enzymatically with IPTs (AtIPT1 and TZS) from dimethylallyl diphosphate and with AMP, ADP, and ATP, respectively, and purified by anion exchange chromatography using a MonoQ column as described previously (8). tZRMP was purchased from Apex Organics (Devon, UK).
Enzyme Assay-Microsomal proteins were incubated in 40 l of reaction mixture (100 mM sodium phosphate, 10% sucrose, iP-type CK, 3 mM NADPH, 1 mg/ml bovine serum albumin, pH 7.5) at 20°C for appropriate periods; the amounts of microsomal protein and molecular species of iP-type CK are indicated in the text. The reaction was terminated by the addition of 400 l of termination buffer (50 mM CHES-NaOH, 0.5 mM MgCl 2 , pH 10.0). The mixture was treated with calf intestine alkaline phosphatase (10 units; Wako Pure Chemical, Osaka, Japan) at 37°C for 30 min. After the addition of 20 l of 20% acetic acid and 2 pmol [ 2 H 5 ]tZR (OlChemim, Olomouc, Czech Republic) to determine the recovery in the following steps, the mixture was centrifuged at 20,000 ϫ g for 20 min. The supernatant was passed through a reversephase column (Oasis HLB, 10 mg/1cc; Waters) pre-equilibrated with 1 ml of methanol and 1 ml of water. The column was washed twice with 1 ml of water, and CKs were eluted with 1 ml of methanol. The solvent was evaporated under vacuum, and residual materials were dissolved in 0.1% acetic acid. After centrifugation, CKs in the supernatant were quantified with LC-MS.
Identification of the Reaction Products as the Nucleotide Forms-Microsomal proteins (100 g) were incubated in 200 l of the reaction mixture with 50 M iPRMP at 20°C for 90 min. The reaction was stopped by adding 20 l of 20% acetic acid. After centrifugation, the supernatant was passed through an Ultrafree MC centrifugal filter (Biomax-10; Millipore). The filtrate was loaded onto an ODS column (Supersphere RP-select B, 4 mm ϫ 250 mm; Merck) by an HPLC system (model 600/717plus/PDA996; Waters) as described previously (23). The fraction containing the reaction product was dried and redissolved in 0.1% acetic acid. After centrifugation, the supernatant was analyzed with a LC-MS system as described above. The mass spectra were obtained by electrospray ionization scanning positive ions from m/z 120 to 500. The capillary voltage was 4 kV.
Extraction of RNA-A. thaliana ecotype Columbia was grown hydroponically on MGRL culture medium (24) for 4 weeks as described (25), and the rosette leaves, flowers, roots, and stems were harvested separately. The organs were soaked in water or 5 M iP by vacuum infiltration and incubated for 1 h. For phytohormone treatment, Arabidopsis seedlings that had been grown on MGRL vertical agar plates for 12 days were sprayed with 5 M of various hormones. After 1 h, total RNA was extracted with an RNeasy plant mini kit (Qiagen).

Isolation of Arabidopsis P450 Genes Involved in tZ
Biosynthesis Using the AtIPT4/P450 Co-expression System-To identify gene(s) that catalyze tZ biosynthesis, we designed a screening method using an AtIPT4/P450 co-expression system in yeast. When AtIPT4, which catalyzes an initial step of the biosynthesis of iP nucleotides, was expressed in yeast, iP and iPR but not tZ and tZR were excreted into the culture medium (Table I, pYES-lacZ/pESC-ATR1-IPT4). Thus, AtIPT4 can synthesize iP nucleotides in yeast cells, whereas authentic yeast enzymes transform iP nucleotides into the corresponding nucleosides and free bases. However, the hydroxylation activity required for tZ production is lacking. We reasoned that if the appropriate P450 were co-expressed with AtIPT4 in yeast, tZtype CKs could be expected to be synthesized. To construct such cell lines, we used two expression vectors, one carrying AtIPT4 and the other carrying one of a series of Arabidopsis P450s. To gain electrotransfer efficiency from NADPH to the plant P450s in yeast, an Arabidopsis NADPH-P450 reductase ATR1 was also expressed in the yeast cells in all cases (19).
CYP735A1 and CYP735A2 (formerly named CYP709A1 and CYP709A2, respectively) are members of the CYP735A subfamily in Arabidopsis and share 79% identity at the amino acid level. CYP735A2 was first described as a CK-inducible gene by a DNA microarray analysis (30). In a phylogenetic tree of Arabidopsis P450 genes, the CYP709B subfamily was a sister

trans-Zeatin Biosynthesis in Arabidopsis
group of the CYP735A subfamily (biobase.dk/P450/p450.shtml). It consisted of three genes, CYP709B1 (At2g46960), CYP-709B2 (At2g46950), and CYP709B3 (At4g27710), which showed 35-40% identity with the CYP735As at the amino acid level. We tested the ability of the CYP709Bs to hydroxylate CKs in the yeast system, but no secretion of tZ-type CKs was detected (data not shown). CYP735A1 and CYP735A2 Utilize iP Nucleotides as Substrates-Microsomal fractions prepared from yeast co-expressing ATR1 and CYP735A1 or CYP735A2 were used to measure the enzymatic activity in vitro. When iP nucleotides were used as substrates, efficient conversion to tZ-type CKs was detected ( Fig. 2A). iPRMP and iPRDP were processed more efficiently than iPRTP by both CYP735As. The conversion efficiency of iP and iPR was less than 10% of that of iPRTP ( Fig. 2A). These results strongly suggest that the CYP735As function as iP nucleotide hydroxylases in vivo.
Although the activity was almost abolished when NADPH was absent from the reaction mixture, the presence of NADH partially restore the activity (Table II). This suggests that the reaction was not completely dependent on NADPH-P450 reductase, and NADH could act as an electron donor probably via other reduction systems such as a NADH-cytochrome b 5 reductase/cytochrome b 5 system (32). The substantial inhibition of the reaction by metyrapone was coincided with the sensitivity to this compound of the CK hydroxylase detected in the microsomal fraction of cauliflower (13). CYP735A1 was more sensitive to metyrapone than CYP735A2 (Table II).
In our routine method of measuring P450 activity, we dephosphorylated the primary products with phosphatase and monitored the nucleosides because of the difficulties to separate nucleotides on reverse-phase columns and the low sensitivity for nucleotides of our MS analysis. To confirm the nature of the primary product, the reaction products generated by CYP735A1 with iPRMP as a substrate were prepared on a large scale and were purified. The retention time determined by HPLC (Fig. 2B) and the spectrum obtained by LC-MS (Fig.  2C) showed that the primary product was tZRMP. We obtained a similar result with CYP735A2 (data not shown).
CYP735A1 and CYP735A2 Catalyze trans-Hydroxylation in Vitro-In the culture medium of yeasts co-expressing AtIPT4/ CYP735A, cZ-type and DZ-type CKs were also detected (Table  I). To establish whether CYP735As catalyze the formation of cis-isomers and/or the reduction of the tZ side chain in vitro, we analyzed the reaction products by LC-MS. When iPRMP was used as substrate, both CYP735As mainly produced the transisomer; production of the cis-isomer occurred at a negligible level (Fig. 3, m/z 352). Although a small peak was detected at m/z 354, it was an unrelated peak because the retention time of the peak (13.3 min) did not coincide with that of the DZR standard (13.1 min; Fig. 3, m/z 354). When tZRMP was incubated with total soluble proteins prepared from the yeast transformants, production of DZR 5Ј-monophosphate was detected. However, this conversion was not CYP735A-dependent (data not shown). We concluded that the production of cZ and DZtype CKs in the yeast expression system was due to endogenous enzymatic activities of the yeast cells and that CYP735As catalyze the stereo-specific formation of tZ nucleotides.
Kinetic Parameters of CYP735A1 and CYP735A2-The contents of P450s in the microsomes were estimated to be 0.062 nmol/mg of microsomal protein for CYP735A1 by analysis of CO difference spectra (Fig. 4). The content of CYP735A2 could not be determined because of its low expression level in yeast (data not shown). We determined the kinetic parameters of CYP735A1 and CYP735A2 for the iP nucleotides (Table III). Of all substrates tested, the K m values of CYP735A1 and CYP735A2 for iPRMP were the smallest, whereas those for iPRTP were the largest. CYP735A2 exhibited higher substrate affinities than CYP735A1 in all cases. The K m values for iP and iPR could not be determined because of the low activities with these substrates. Furthermore, comparisons of the catalytic  a Microsomal fractions were prepared from yeasts harboring pESC-ATR1/pYES-CYP735As.
b Activity was measured in the reaction mixture with 10 g of microsomal protein and 10 M iPRMP. The activities of CYP735A1 and CYP735A2 in control were 21.4 Ϯ 0.5 and 0.57 Ϯ 0.03 pmol/min, respectively (n ϭ 3).
c Relative activity with the value in control taken as 100. d Activity was measured without NADPH. e Activity was measured with 3 mM NADH. f 1% Me 2 SO was added to the reaction mixture. g 1% Me 2 SO and metyrapone were added to the reaction mixture.

trans-Zeatin Biosynthesis in Arabidopsis
constants (k cat ) and the specificity constants (k cat /K m ) of the reactions strongly suggested that CYP735As preferentially act on iPRMP and iPRDP (Table III). Expression Profile of CYP735As Genes-To evaluate the levels of CYP735As expression in different organs and the responsiveness to exogenous CK, rosette leaves, flowers, roots, and stems of Arabidopsis were incubated in water or iP, and accumulation levels of CYP735A transcripts were analyzed by a quantitative real time PCR method. UBQ10 (33) and ARR5 (34) were used as internal negative and positive controls, respectively. In the plants, the transcripts of CYP735A1 were more abundant in roots and flowers than in other organs (Fig. 5, open  bars). On the other hand, the accumulation of the CYP735A2 transcript was highest in roots and stems (Fig. 5, open bars).
Upon treatment with iP, the transcript accumulation levels were not significantly changed except for that of CYP735A2 in roots (Fig. 5, closed bars). Exogenous iP promoted ARR5 expression in all organs. Thus, the expression of CYP735A1 and CYP735A2 was differentially regulated in terms of organ specificity and CK responsiveness.
iP, which had promoted only the accumulation of CYP735A2 transcripts in roots when applied to older plants (Fig. 5), had a strong positive effect on CYP735A2 mRNA levels and a weaker but significant one on CYP735A1 transcript accumulation in roots of 12-day-old seedlings (Fig. 6). The iP-dependent accumulation of CYP735A transcripts was transient; it reached a maximum after 30 -60 min and decreased to the initial level after 4 h (data not shown). In both genes, a strong enhancement of mRNA accumulation was also observed when tZ instead of iP was applied (Fig. 6). cZ also promoted transcript accumulation, but its effect was statistically significant only in CYP735A2 (Fig. 6). On the other hand, application of auxin or abscisic acid decreased the accumulation of both CYP735A transcripts, suggesting participation of the phytohormones in the regulation of tZ biosynthesis. Gibberellin and brassinolide had no effects. DISCUSSION In this study, we demonstrated that the Arabidopsis CYP735A1 and CYP735A2 encode CK hydroxylases that catalyze tZ biosynthesis via the iPRMP-dependent pathway. In vitro analysis revealed that the stereo-specific trans-hydroxylation occurs at the nucleotide level. So far, hydroxylating activities of iP-type CKs have been examined only at the base (iP) and nucleoside (iPR) levels in microsomal fractions of cauliflower (13). Therefore it is uncertain whether the cauliflower enzyme(s) also act on nucleotides. We attempted to determine CK hydroxylase activity in microsomal fractions prepared from Arabidopsis roots with various iP-type CKs but could not detect any. This was probably due to the low level of CK hydroxylase in the preparation (data not shown).
Although CK bases and nucleosides are regarded as the physiologically active form and the translocation form, respectively (3), the role of the nucleotides remains to be clarified. If the conversion to tZ-type CKs is a nucleotide-specific process in planta, one of the roles of the CK nucleotides might be to form a metabolic pool for the side chain hydroxylation. iP nucleotides are primary products of the CK biosynthetic reaction (6,7), indicating that hydroxylation occurs at an early stage of CK metabolism. Such a metabolic hierarchy might be important to maintain the homeostasis between iP-and tZ-type CKs.
The stereo specificity of the tZ formation catalyzed by CYP735As indicates that an unknown system must be involved in cis-hydroxylation. This is also supported by analyses using 13 C-labeled precursors specific to the MEP or MVA pathways, which demonstrated distinct isoprenoid sources in the biosyntheses of tZ and cZ (17). At present, it is thought that the hydroxylation of the prenyl moiety of iPR residues that are contained in some tRNA species, and their subsequent release by tRNA degradation is a major pathway for cZ biosynthesis (35) (Fig. 1). miaE is a cZ biosynthesis gene in Salmonella typhimurium (36); so far, no miaE homologues have been found in plant genomes.
The fact that CYP735As and IPTs are differentially regulated by CKs and auxin is intriguing. In Arabidopsis, the ac- FIG. 4. CO difference spectra of recombinant CYP735As expressed in yeast. Microsomal proteins prepared from yeast co-expressing CYP735A1 and ATR1 (CYP735A1) or lacZ and ATR1 (Control) were used. CO reduced difference spectra were recorded using the yeast microsomes (2.0 mg protein/ml) in a buffer containing 100 mM potassium phosphate (pH 7.5) and 20% glycerol.

trans-Zeatin Biosynthesis in Arabidopsis
cumulation of the transcripts of AtIPT5 and AtIPT7 is positively regulated by auxin, whereas the transcript levels of AtIPT1, AtPT3, AtIPT5, and AtIPT7 are negatively regulated by CK (37). The transcripts of the two CYP735As behaved in a manner opposite to those of AtIPT5 and AtIPT7 (Figs. 5 and 6). Genes for cytokinin oxidase/dehydrogenase (CKX), a CK degradation enzyme, are also known to be up-regulated by CKs (38). These regulation patterns suggest that the genes mentioned are involved as antagonists in the regulation of the cellular CK level and of the balance between iP-and tZ-type CKs that interact with auxin.
Very recently, it has been reported that auxin also negatively regulated tZ biosynthesis via the iPRMP-independent pathway (39). Treatment of Arabidopsis seedlings with 1-naphthalenacetic acid greatly reduced the accumulation of tZRMP but not that of iPRMP (39). Although biochemical nature of the iPRMP-independent pathway has not been elucidated, these lines of evidence suggest that auxin is an important factor that regulates the both iPRMP-dependent and iPRMP-independent pathways.
Little is known about the physiological significance of CK side chain hydroxylation. In other phytohormones, hydroxylation reactions catalyzed by P450s participate in biosynthesis (40 -43), inactivation (44), and catabolism (45,46). Hydroxylation is not essential for CK catabolism because CKX cleaves side chains from iP as well as tZ (47) (Fig. 1). The modification might be involved in reversible inactivations of CKs by CK  O-glucosyltransferases (3, 48). The O-glucosides are biologically inactive, insensitive to CKX, and reversibly converted to the active form by ␤-glucosidase (49) (Fig. 1). CK-dependent induction of CYP735A expression (Figs. 5 and 6) might cause a change in the metabolic flow that leads to storage of excess CKs in an inactive form. It would be interesting to see whether genes for the O-glucosyltransferases are also regulated by CKs.
The trans-hydroxylation might be involved also in the regulation of the long range translocation of CKs from roots to shoots via the xylem. This transport plays a role in signaling of nitrogen nutrition (23), adventitious root formation (50), and shoot branching (51). tZ-type CKs such as tZR are the dominant CKs in xylem exudates (23,52). Therefore, predominant expression of CYP735As in roots and root-specific induction of CYP735A2 by CK (Fig. 5) appears consistent with the idea. On the other hand, leaf exudates contained mainly iP-type CKs (53). Thus, trans-hydroxylation may be important for the compartmentalization of the CK species and may control the direction of translocation.